rm(list = ls())
library(Seurat)
library(dplyr)
library(patchwork)
library(ggplot2)
load( "sce.Rdata")
scRNA = sce
scRNA@meta.data$celltype = Idents(scRNA)
ctys = levels(scRNA)
ctys

## [1] "naive B"     "CD8 T"       "Naive CD4 T" "plasma B"    "CD14+ Mono" 
## [6] "endothelial" "Fibroblasts" "NK"          "DC"

scRNA.markers <- FindAllMarkers(scRNA, min.pct = 0.25, 
                    logfc.threshold = 0.25)
head(scRNA.markers)

##                  p_val avg_log2FC pct.1 pct.2     p_val_adj cluster     gene
## CD79A     0.000000e+00   5.356611 0.915 0.060  0.000000e+00 naive B    CD79A
## BANK1     0.000000e+00   6.778892 0.853 0.026  0.000000e+00 naive B    BANK1
## MS4A1     0.000000e+00   5.677155 0.848 0.050  0.000000e+00 naive B    MS4A1
## HLA-DRA  1.482079e-306   2.775984 0.994 0.336 3.060197e-302 naive B  HLA-DRA
## HLA-DQB1 7.341732e-295   2.622558 0.926 0.207 1.515921e-290 naive B HLA-DQB1
## HLA-DQA1 5.404715e-292   2.727477 0.865 0.138 1.115966e-287 naive B HLA-DQA1

colnames(scRNA.markers)[6] = "celltype"
k = scRNA.markers$p_val_adj<0.05;table(k)

## k
## FALSE  TRUE 
##  3960  8946

scRNA.markers = scRNA.markers[k,]

#上下调
scRNA.markers$label <- ifelse(scRNA.markers$avg_log2FC<0,"sigDown","sigUp")
topgene <- scRNA.markers %>%
  group_by(celltype) %>%
  top_n(n = 10, wt = avg_log2FC) %>%
  bind_rows(group_by(scRNA.markers, celltype) %>%
              top_n(n = 10, wt = -avg_log2FC))
head(topgene)

## # A tibble: 6 × 8
## # Groups:   celltype [1]
##       p_val avg_log2FC pct.1 pct.2 p_val_adj celltype gene      label
##       <dbl>      <dbl> <dbl> <dbl>     <dbl> <fct>    <chr>     <chr>
## 1 0               6.78 0.853 0.026 0         naive B  BANK1     sigUp
## 2 3.13e-250       6.98 0.498 0.011 6.47e-246 naive B  LINC00926 sigUp
## 3 2.59e-207       6.02 0.423 0.01  5.34e-203 naive B  CD24      sigUp
## 4 1.42e-204       6.90 0.411 0.008 2.92e-200 naive B  LINC02397 sigUp
## 5 1.15e-196       7.13 0.403 0.01  2.38e-192 naive B  IGHD      sigUp
## 6 5.77e-166       7.47 0.335 0.006 1.19e-161 naive B  PAX5      sigUp

#根据log2FC范围确定背景柱长度：
dfbar = scRNA.markers %>%
  group_by(celltype) %>%
    summarise(low = round(min(avg_log2FC)-0.5),
            up = round(max(avg_log2FC)+0.5))

#绘制背景柱和散点图：
p1 <- ggplot()+
  geom_col(aes(x = celltype ,y = low),dfbar,
           fill = "#dcdcdc",alpha = 0.6)+
  geom_col(aes(x = celltype ,y = up),dfbar,
           fill = "#dcdcdc",alpha = 0.6)+
  geom_jitter(aes(x = celltype, y = avg_log2FC, color = label),scRNA.markers,
              width =0.4,size = 1)+
  scale_color_manual(values = c("#0077c0","#c72d2e"))+
  theme_classic()
p1

#X轴的色块标签：
library(RColorBrewer)
mycol <- colorRampPalette(rev(brewer.pal(n = 7, name ="Set1")))(length(ctys))
p2 <- p1 + 
  geom_tile(aes(x = ctys,y = 0),
            height = 0.5,fill = mycol, show.legend = F)+
  geom_text(aes(x= ctys, y = 0, label = ctys),
            size = 3,fontface = "bold")
p2

library(ggrepel)
#给每个Cluster差异表达top基因加上标签,调整细节：
p3 <- p2 + 
  geom_text_repel(aes(x = celltype,y = avg_log2FC,label = gene),
                  topgene,size = 3 )+
  labs(x = "CellType",y = "Average log2FoldChange",
       title = "Differential expression genes")+
  theme(
    plot.title = element_text(size = 14,color = "black",face = "bold"),
    axis.title = element_text(size = 12,color = "black",face = "bold"),
    axis.line.y = element_line(color = "black",linewidth = 0.8),
    axis.line.x = element_blank(),
    axis.text.x = element_blank(),
    axis.ticks.x = element_blank(),
    panel.grid = element_blank(),
    legend.position  = c(0.98,0.96),
    legend.background = element_blank(),
    legend.title = element_blank(),
    legend.direction = "vertical",
    legend.justification = c(1,0),
    legend.text = element_text(size = 12)
  )+
  guides(color = guide_legend(override.aes = list(size = 4)))  
p3